Support destructuring in function arguments

[marijnh]

Basically, allow things like fn foo((a, b): (int, int), @u: @something) {}.

We talked about this in the Toronto meeting. I was surprised that no issue existed for it.

[Aluminium95]

Would enum and record support destructuring in function arguments too ?

[brson]

Do we really need this? I know that occasionally I have thought it would be nice to have, but we have lasted this long without anybody complaining loudly about it, so it doesn't seem vital.

[marijnh]

It'd be consistent. I occasionally find myself wanting to use this -- mostly for tuple arguments.

[brson]

I wonder how it would interact with our various argument modes. Does it effectively turn them all into copies? Is it only compatible with certain modes?

[marijnh]

Hm. I assumed that regular modes (non-copy/move) would be able to just create views into the existing value without copying, since such args are immutable locals, but I guess if that value has mutable components, this would mess with the invariant that locals don't get their values changes from under them.

For by-copy and by-move the story gets even more involved. Maybe you're right that we shouldn't bother.

[marijnh]

I'm closing this. If someone feels strongly about supporting this, discuss.